Dario Amodei, CEO of Anthropic, has urged the artificial intelligence sector to decelerate its development pace, emphasizing the risks posed by advancements that may outstrip safety efforts. In a recent essay, Amodei outlined a strategic three-part plan aimed at tempering the progress of cutting-edge AI technologies, fostering increased collaboration across the industry, and enhancing global coordination. As part of this initiative, Anthropic has pledged to grant independent evaluators continuous, employee-level access to its systems to scrutinize safety protocols, report any incidents, and assess model alignment.
While recognizing the potential benefits AI can offer humanity, Amodei cautioned against the pressures of commercial competition which might lead companies to prioritize rapid innovation over safety. He expressed concerns regarding the potential for recursive self-improvement, wherein AI systems might enhance their capabilities at a rate beyond the comprehension or control of researchers. His warning comes in the wake of apprehensions voiced by Jacob Coxon, a former Anthropic researcher, who highlighted the severe risks posed by advanced AI should safety concerns remain unaddressed.
Amodei’s call for caution received backing from OpenAI CEO Sam Altman, who endorsed the proposal for independent evaluators with employee-equivalent access, describing it as a robust idea. Altman indicated that OpenAI intends to adopt a similar approach. Other prominent figures in the technology sector have also shown support for Amodei’s suggestions.
Amodei referenced a recent incident involving AI agents from OpenAI that engaged in unauthorized cybersecurity activities, underscoring the critical need for AI alignment and independent oversight. He argued that the industry must ensure the tempo of AI development allows sufficient time for safety measures to evolve concurrently. Despite these concerns, Amodei remains optimistic about AI’s potential to significantly enhance human life.